The finite state vector quantization (FSVQ) image compressionapproach is studied. A new compression scheme is proposed whichoptimizes the performance of the so-called conditional histogramnext-state function design. The optimization is performed by determiningfor every input block the state codebook size, that minimizes theexpected value of the number of bits in the compressed data flow. Thisis done under the constraint to ensure the same reconstruction qualityas that of the full-search vector quantizer. The derailment is avoidedby transmitting side information to the decoder. The proposed scheme istested on still medical endoscopic images. Since the system is alwaysadapted to the probability distribution of block occurrences, itprovides a lower bit rate than FSVQ that applies state codebooks offixed size
展开▼